Implementation of the paper: StyleBank: An Explicit Representation for Neural Image Style Transfer
- Training
- Incremental Learning
- Prepare your content image dataset (I use MS-COCO)
- Prepare your style image dataset
It takes about 1 days to train on a GeForce GTX 1080 Ti. Pre-trained weights are provided. Just load the encoder, decoder and the stylebank you want.
| Style | Source | Transfered |
|---|---|---|
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
 |
 |
 |
 |
 |